Little is known about the effects of additional survey language(s) on representation bias in general population surveys. Up to now, language(s) of the population as well as logistic and financial reasons have been the main drivers for deciding which language(s) are offered to the sample members. However, effects of an additional language are not clear and may depend on the complementary sample addressed by this language, characteristics of the survey, as well as their complex interaction.
In this article, we present a framework to investigate the potential of additional survey languages to reduce representation bias by focusing on who among the population can be additionally covered. In particular, we investigate the effects of adding survey languages on coverage bias across different dimensions: 1) languages with different potentials to add different groups of the complementary population 2) variables defining different population groups across whom bias is measured, 3) language mastery requirements to complete the survey for sample members with 4) different socio-demographic compositions. Results from a potential study using data from Switzerland show that the choice of an additional language depends primarily on the variable defining different population groups , then on the language mastery necessary, and least so on the socio-demographic composition of the sample.
Presented by:
Oliver Lipps, University of Lausanne
Date & time:
January 22, 2018 4:00 pm - January 22, 2018 5:30 pm
Venue:
ISER Large Seminar Room, 2N2 4.16
External seminars home